Builds / RTX 2060 Super 8GB
RTX 2060 Super 8 GB + Core i9-10900K: Qwen3.5 9B Q4_K_M at 54.4 tok/s, Gemma 4 12B at 27.7 (llama.cpp)
Last updated:
Source: mustafa.net ↗📏 How measured: reported by the original source; RigBase re-checked the numbers at that source
- GPU(s)
- RTX 2060 Super 8GB
Amazon · eBay (used) · Newegg - VRAM / unified memory (GB)
- 8
- CPU
- Intel Core i9-10900K
Amazon · eBay (used) · Newegg - Motherboard
- not stated
- RAM
- not stated
- PSU
- not stated
- Other
- Windows 11 + Docker Desktop (Linux layer, may cost a little speed); all layers on the card
- Reported total cost
- not stated
- Power
- not stated
Measured speed (generation tokens/sec)
| Model | Quant | Engine | tok/s | Prompt processing (tok/s) | Details |
|---|---|---|---|---|---|
| Qwen3.5 9B | Q4_K_M | llama.cpp server (CUDA, Docker) | 54.4 | not stated | context: 8,192 ctx, short fixed prompt📏 llama.cpp server (Docker, CUDA image), 8,192 ctx, 256 tokens at temperature 0; middle of three runs after a warm-up (runs agreed within ~1%); 5.5 GB card memory in use |
| Gemma 4 12B | Q4_K_M | llama.cpp server (CUDA, Docker), forced full offload | 27.7 | not stated | context: 8,192 ctx, short fixed prompt📏 llama.cpp server (Docker, CUDA image), 8,192 ctx, 256 tokens at temperature 0; middle of three runs after a warm-up (runs agreed within ~1%). 7.1 GB file forced fully onto the 8 GB card (card memory full); with llama.cpp's automatic placement some layers stayed in system RAM and it ran at 11.3 tok/s |
| Qwen3.5 4B legacy | Q4_K_M | llama.cpp server (CUDA, Docker) | 83.7 | not stated | context: 8,192 ctx, short fixed prompt📏 llama.cpp server (Docker, CUDA image), 8,192 ctx, 256 tokens at temperature 0; middle of three runs after a warm-up (runs agreed within ~1%); 3.3 GB card memory in use |
Mustafa.net homelab test, measured 12 Oct 2026 (published 13 Oct 2026) to check the site's own Will It Run? calculator. A 2019 8 GB card still runs current 4B-12B models at Q4. Same files and prompt as the Mac mini M4 build from the same article. RAM, motherboard and PSU not stated.
Source: https://mustafa.net/2026/10/13/local-llm-calculator-tested-four-machines/
Blank fields = not stated in the source. We never estimate build data; GPU street prices above are separately sourced and dated.
Embed this result
Copy into a README or blog. Links back to this build. See badge docs (CC BY 4.0 attribution). The SVG does not invent speeds — numbers stay on the build page.
Markdown
[](https://getrigbase.com/builds/rtx-2060-super-8gb-i9-10900k-qwen35-gemma4-mustafa/)
HTML
<a href="https://getrigbase.com/builds/rtx-2060-super-8gb-i9-10900k-qwen35-gemma4-mustafa/" rel="noopener"><img src="https://getrigbase.com/badge/rtx-2060-super-8gb-i9-10900k-qwen35-gemma4-mustafa.svg" alt="Measured on RigBase" height="20"/></a>
Preview: 
· How to use
Rent before you buy: try this model on a cloud GPU for a few dollars first: Runpod (referral link; sign up with Google to get the new-user bonus)